Regional inequality in human capital formation in Europe, 1790–1880

Abstract

Recent theoretical advances reveal the importance of human capital for long-run economic growth. However, the absence of data makes it difficult to measure human capital before 1870 at the national level, let alone at the regional level within countries. By using the age heaping method and a large, new data set, we approximate the numeracy values in more than 570 regions in Europe between 1790 and 1880. The results indicate a significant gap in numeracy levels between advanced west and central European countries and the rest of Europe. Nevertheless, differences in basic numeracy between and within countries became smaller over the nineteenth century, as the periphery solved its basic numeracy problem.

Keywords:

• human capital
• numeracy
• regional development
• inequality
• Europe

JEL Classification:

• N33
• N93
• O18

Notes

¹Galor, ‘Demographic Transition’ (2012). Galor, ‘Unified’ (2005). Galor/Moav, ‘Selection’ (2002). Galor/Weil, ‘Population’ (2000).

²Romer, ‘Endogenous’ (1990). Lucas, ‘Mechanics’ (1988).

³For example, Allen, British Industrial (2009). Becker/Woessmann, ‘Wrong’ (2009).

⁴Cipolla, Literacy (1969).

⁵Krugman, Geography (1991).

⁶For example, Crayen/Baten, ‘Global’ (2010). A'Hearn et al., ‘Quantifying’ (2009).

⁷O'Rourke/Williamson, ‘Periphery’ (1997).

⁸Good/Ma, ‘Perspective’ (1999).

⁹Ivanov/Tooze, ‘Convergence’ (2007).

¹⁰Foreman-Peck/Lains, ‘Periphery’ (2000).

¹¹Tortella, ‘Patterns’ (1994).

¹²Sandberg, ‘Impoverished’ (1979).

¹³See, e.g., Becker et al., ‘Catch-up’ (2011). Zamagni, ‘Istruzione’ (1978).

¹⁴As an example, O'Rourke and Williamson argue that, for the period between 1870 and 1913, forces of globalisation were more important for economic development than education. See O'Rourke/Williamson, ‘Periphery’ (1997).

¹⁵For an excellent overview see Demeulemeester/Diebolt, ‘Links’ (2011).

¹⁶Mitch, ‘Role’ (1993).

¹⁷Mokyr, Gifts (2002).

¹⁸Baten/van Zanden, ‘Book production’ (2008).

¹⁹Mokyr, Gifts (2002), 3.

²⁰Allen, British Industrial (2009).

²¹Becker/Woessmann, ‘Wrong’ (2009).

²²Tollnek/Baten, ‘The Farmer's U’ (2011).

²³For example, Becker/Woessmann, ‘Investment’ (2012). A'Hearn et al., ‘Quantifying’ (2009). Baten/van Zanden, ‘Book production’ (2008). Benavot/Riddle, ‘Expansion’ (1988).

²⁴Brint, Schools (2006).

²⁵Soysal/Strang, ‘Construction’ (1989).

²⁶Vincent, Rise (2010).

²⁷Soysal/Strang, ‘Construction’ (1989).

²⁸Weber, Protestant (1958).

²⁹Becker/Woessmann, ‘Wrong’ (2009).

³⁰These dates refer to the years when the first important general schooling laws were passed. These laws aimed at enforcing the attendance of all children until they reached some specific age.

³¹Flora, Indikatoren (1975).

³²Vincent, Rise (2010). Adick, ‘Trends’ (2003).

³³Schneider, ‘Bildungsentwicklung’ (1982).

³⁴A notable exception is Prussia [see, e.g., Becker/Woessmann, ‘Wrong’ (2009)].

³⁵For example, Reis, ‘Economic Growth’ (2005). Mitch, ‘Role’ (1993). Schofield, ‘Dimensions’ (1981).

³⁶Cipolla, Literacy (1969).

³⁷Earlier studies on literacy include, e.g., the one by Fleury/Valmary, Progrès (1957).

³⁸Graff, Legacies (1987), Graff, Literacy Myth (1991). Graff et al., Understanding (2009).

³⁹Allen, British Industrial (2009). See also Graff, Legacies (1987). Cressy, Literacy (1980).

⁴⁰A'Hearn et al., ‘Quantifying’ (2009).

⁴¹Crayen/Baten, ‘Global’ (2010).

⁴²For example, A'Hearn et al., ‘Quantifying’ (2009). Manzel/Baten, ‘Gender Equality’ (2009). de Moor/van Zanden, ‘Benadering’ (2008).

⁴³Goyer/Draaijer, Handbook (1992), 6.

⁴⁴Goyer/Draaijer, Handbook (1992).

⁴⁵Sahai, ‘Census’ (1988).

⁴⁶Yet, age heaping is still observable in current censuses in parts of Asia and Africa.

⁴⁷The question may arise whether all individuals stated their ages by themselves. Friesen et al. find by distinguishing between males and females in their data that a large gender gap in literacy existed also in numeracy. This is a strong indication that females were asked themselves because the differences between both gender gaps would be much more pronounced. Friesen et al., ‘Women’ (2012).

⁴⁸A'Hearn et al., ‘Quantifying’ (2009).

⁴⁹Moreover, because this has been the standard way of calculating numeracy values derived from age heaping, it allows to be in line with the other contributions. This standardisation enables us and other researchers later on to compare the numeracy values from different studies directly.

⁵⁰Crayen/Baten, ‘Global’ (2010).

⁵¹For example, Crayen/Baten, ‘Global’ (2010).

⁵²Nevertheless, we want to indicate that the possibility remains that numeracy could improve in a population independently of improved literacy or schooling. Baten et al. consider China and find evidence for such a case in this East Asian country. Baten et al., ‘China’ (2010). Basic numeracy is in fact an ability which can also be trained in households by using calendars, numerically demanding games for children and similar devices.

⁵³Yet if they did so, then age heaping would normally be equal to zero, which can mostly be identified.

⁵⁴Values below 100 are also possible but are normally found in samples with a low number of observations.

⁵⁵ABCC refers to the initials of A'Hearn, Baten and Crayen (of A'Hearn et al., ‘Quantifying’ (2009)) and Gregory Clark, who suggested the index in a comment.

⁵⁶Crayen/Baten, ‘Global’ (2010). In practice, the precise adjustment of the corresponding age group is an addition of “0.2 Whipple units for every Whipple unit above 100”, Crayen/Baten, ‘Global’ (2010), 95. This implies a reduction of ABCC estimates by roughly one quarter.

⁵⁷Crayen/Baten, ‘Global’ (2010).

⁵⁸Except for the census of Greece in 1907, where it is only possible to include age data from people between the ages of 23 and 32 because data for individual years are only available up to the age of 34.

⁵⁹The exact sources are documented in the Appendix.

⁶⁰In contrast, data on the population of the major cities were collected by census takers, as in other European countries. Thus, these data are also characterised by age heaping. Nevertheless, because all our other data are homogeneous in the sense that they always cover the whole population, and because the exact relationship between the urban and rural population is not yet sufficiently clear, we decided not to include these data in the framework of this study.

⁶¹Nevertheless, the Finnish data already indicate that numeracy should be at a high level in this country.

⁶²For Romania, we only have data for the parts of the Hungarian Kingdom.

⁶³These are Belgium, Bulgaria, Czech Republic, Denmark, Germany, Ireland, Greece, Spain, France, Italy, Cyprus, Latvia, Lithuania, Luxembourg, Hungary, Malta, the Netherlands, Austria, Poland, Portugal, Romania, Slovenia, Slovakia, Finland, Sweden and the UK.

⁶⁴Lichtenstein, Norway and Switzerland.

⁶⁵Montenegro, Croatia, FYROM, Turkey and Iceland.

⁶⁶Notable exceptions include Spain and France.

⁶⁷Martí-Henneberg, ‘Concentration’ (2005).

⁶⁸As a caveat, we should note that we cannot directly compare the historical numeracy and today's numeracy level due to population movements, but this comparison is the purpose of this study.

⁶⁹For mapping purposes, we decided to use the smallest available territorial unit to highlight the regional differences as clearly as possible.

⁷⁰Or similarly small units for countries not included in the NUTS classification.

⁷¹Our classification of the core countries corresponds to the one employed by O'Rourke and Williamson, except that it has been extended by also including Luxembourg. See O'Rourke/Williamson, ‘Periphery’ (1997).

⁷²Core Europe is comprised of Belgium, Switzerland, Germany, France, Luxembourg, the Netherlands and the United Kingdom, Austria-Hungary of Austria, Czech Republic, Croatia, Hungary, Romania, Slovenia and Slovakia, the Western Periphery of Denmark, Spain, Ireland, Iceland, Italy, Norway and Portugal, Eastern Europe of Belarus, Lithuania, Latvia, Estonia, Moldova, Poland, Russia and Ukraine and South-East Europe of Albania, Armenia, Azerbaijan, Bosnia-Hercegovina, Bulgaria, Cyprus, Georgia, Greece, Macedonia and Serbia.

⁷³Today's Croatia was split between Cisleithania (mostly Dalmatia) and Transleithania (Croatia-Slavonia). Here we refer only to the part belonging to Transleithania.

⁷⁴Allen, British Industrial (2009).

⁷⁵However, there might be a bias in signature rates because we are considering all of Germany, whereas signature rates typically come from few locations. Depending on the location this may explain the relatively low literacy values for Germany at that time.

⁷⁶Giedraitiene et al., ‘Lithuania’ (2007).

⁷⁷Sroka, ‘Belarus’ (2007).

⁷⁸Mironov, ‘Development’ (1991).

⁷⁹Raun, ‘Movements’ (1986).

⁸⁰Particularly important was the Church Law of 1686.

⁸¹Johansson, ‘History’ (2009).

⁸²The most important exception to this trend is the Caucasus region, which has equally little numeracy.

⁸³In particular, this is Serbia (independence fully recognised internationally in 1878), Bosnia-Herzegovina (occupied by the Austro-Hungarian Empire in 1878 and later annexed in 1908), Montenegro (independent in 1878), Albania, Bulgaria (self-government re-established in 1878 (Principality of Bulgaria), unification with Eastern Rumelia in 1885, de jure independence in 1908), Romania (independence in 1878), Macedonia (part of Serbia in 1913) and to some extent Greece (already independent in 1830).

⁸⁴The principality of Bulgaria was already de facto independent in 1878 and was unified with Eastern Rumelia in 1885.

⁸⁵Daskalova, Literacy (1996), 6.

⁸⁶Crampton, Bulgaria (2007).

⁸⁷Crampton, Bulgaria (2007).

⁸⁸Daskalova, Literacy (1996).

⁸⁹Kostovicova, Kosovo (2005).

⁹⁰In order to discern as many regional differences as possible, the lowest available administrative classification (NUTS 3 or NUTS 2) has been selected in the cartographic representation.

⁹¹Ozouf/Furet, Lire (1977).

⁹²Guttormsson, ‘Literacy’ (1990).

⁹³For example, Lockridge, Literacy (1974).

⁹⁴Houston, ‘Scotland’ (1987), 60.

⁹⁵Nevertheless, the reasons why the Lowlands of Scotland do not generally show higher numeracy rates than many regions in England still have to be explored further.

⁹⁶Núnez, Fuente (1992).

⁹⁷The numeracy of the surrounding region is shown in the maps; the numeracy of the city itself is still higher throughout time. We refer here to the numeracy of the city.

⁹⁸Odessa had higher literacy rates than St. Petersburg but lower rates than Moscow. See Herlihy, Odessa (1986).

⁹⁹Goodwin/Grennes, ‘Tsarist Russia’ (1998).

¹⁰⁰Still, one has to take into consideration the already high level of attained numeracy in Prussia. For this reason, the differences are not very large.

¹⁰¹Not all countries are included for the calculation of the ABCC means since only data on the national level are available for the smaller countries (e.g. Luxembourg, Iceland, Estonia, Cyprus) and thus regional inequalities cannot be measured.

¹⁰²The underlying census was the first undertaken in Yugoslavia in 1921.